arXiv subjects

Shijie Liu

Publications and source records attributed to Shijie Liu.

31 records · Page 2Linked to original sources

arXiv2023

It's Simplex! Disaggregating Measures to Improve Certified Robustness

Certified robustness circumvents the fragility of defences against adversarial attacks, by endowing model predictions with guarantees of class invariance for attacks up to a calculated size. While there is value in these certifications, the techniques through which we assess their performance do not present a proper accounting of their strengths and weaknesses, as their analysis has eschewed consideration of performance over individual samples in favour of aggregated measures. By considering the potential output space of certified models, this work presents two distinct approaches to improve the analysis of certification mechanisms, that allow for both dataset-independent and dataset-dependent measures of certification performance. Embracing such a perspective uncovers new certification approaches, which have the potential to more than double the achievable radius of certification, relative to current state-of-the-art. Empirical evaluation verifies that our new approach can certify $9\%$ more samples at noise scale $σ= 1$, with greater relative improvements observed as the difficulty of the predictive task increases.

cs.LG↗

arXiv2023

MEMD-ABSA: A Multi-Element Multi-Domain Dataset for Aspect-Based Sentiment Analysis

Aspect-based sentiment analysis is a long-standing research interest in the field of opinion mining, and in recent years, researchers have gradually shifted their focus from simple ABSA subtasks to end-to-end multi-element ABSA tasks. However, the datasets currently used in the research are limited to individual elements of specific tasks, usually focusing on in-domain settings, ignoring implicit aspects and opinions, and with a small data scale. To address these issues, we propose a large-scale Multi-Element Multi-Domain dataset (MEMD) that covers the four elements across five domains, including nearly 20,000 review sentences and 30,000 quadruples annotated with explicit and implicit aspects and opinions for ABSA research. Meanwhile, we evaluate generative and non-generative baselines on multiple ABSA subtasks under the open domain setting, and the results show that open domain ABSA as well as mining implicit aspects and opinions remain ongoing challenges to be addressed. The datasets are publicly released at \url{https://github.com/NUSTM/MEMD-ABSA}.

cs.CL↗

arXiv2023

Dual Residual Attention Network for Image Denoising

In image denoising, deep convolutional neural networks (CNNs) can obtain favorable performance on removing spatially invariant noise. However, many of these networks cannot perform well on removing the real noise (i.e. spatially variant noise) generated during image acquisition or transmission, which severely sets back their application in practical image denoising tasks. Instead of continuously increasing the network depth, many researchers have revealed that expanding the width of networks can also be a useful way to improve model performance. It also has been verified that feature filtering can promote the learning ability of the models. Therefore, in this paper, we propose a novel Dual-branch Residual Attention Network (DRANet) for image denoising, which has both the merits of a wide model architecture and attention-guided feature learning. The proposed DRANet includes two different parallel branches, which can capture complementary features to enhance the learning ability of the model. We designed a new residual attention block (RAB) and a novel hybrid dilated residual attention block (HDRAB) for the upper and the lower branches, respectively. The RAB and HDRAB can capture rich local features through multiple skip connections between different convolutional layers, and the unimportant features are dropped by the residual attention modules. Meanwhile, the long skip connections in each branch, and the global feature fusion between the two parallel branches can capture the global features as well. Moreover, the proposed DRANet uses downsampling operations and dilated convolutions to increase the size of the receptive field, which can enable DRANet to capture more image context information. Extensive experiments demonstrate that compared with other state-of-the-art denoising methods, our DRANet can produce competitive denoising performance both on synthetic and real-world noise removal.

eess.IV↗

arXiv2022

Double Bubble, Toil and Trouble: Enhancing Certified Robustness through Transitivity

In response to subtle adversarial examples flipping classifications of neural network models, recent research has promoted certified robustness as a solution. There, invariance of predictions to all norm-bounded attacks is achieved through randomised smoothing of network inputs. Today's state-of-the-art certifications make optimal use of the class output scores at the input instance under test: no better radius of certification (under the $L_2$ norm) is possible given only these score. However, it is an open question as to whether such lower bounds can be improved using local information around the instance under test. In this work, we demonstrate how today's "optimal" certificates can be improved by exploiting both the transitivity of certifications, and the geometry of the input space, giving rise to what we term Geometrically-Informed Certified Robustness. By considering the smallest distance to points on the boundary of a set of certifications this approach improves certifications for more than $80\%$ of Tiny-Imagenet instances, yielding an on average $5 \%$ increase in the associated certification. When incorporating training time processes that enhance the certified radius, our technique shows even more promising results, with a uniform $4$ percentage point increase in the achieved certified radius.

cs.LG↗

arXiv2021

Transformer for Polyp Detection

In recent years, as the Transformer has performed increasingly well on NLP tasks, many researchers have ported the Transformer structure to vision tasks ,bridging the gap between NLP and CV tasks. In this work, we evaluate some deep learning network for the detection track. Because the ground truth is mask, so we can try both the current detection and segmentation method. We select the DETR as our baseline through experiment. Besides, we modify the train strategy to fit the dataset.

eess.IV↗

arXiv2021

Observation of flat-band and band transition in the synthetic space

Constructions of synthetic lattices in photonics attract growingly attentions for exploring interesting physics beyond the geometric dimensionality, among which modulated ring resonator system has been proved as a powerful platform to create different kinds of connectivities between resonant modes along the synthetic frequency dimension with many theoretical proposals. Various experimental realizations are investigated in a single ring resonator, while building beyond simple synthetic lattices in multiple rings with different types remains lacking, which desires to be accomplished as an important step further. Here, we implement the experimental demonstration of generating the one-dimensional Lieb lattice along the frequency axis of light, realized in two coupled ring resonators while the larger ring undergoing dynamic modulation. Such synthetic photonic structure naturally exhibits the physics of flat band. We show that the time-resolved band structure read out from the drop-port output of the excited ring is the intensity projection of the band structure onto specific resonant mode in the synthetic momentum space, where gapless flat band, mode localization effect, and flat to non-flat band transition are observed in experiments and verified by simulations. Our work gives a direct evidence for the constructing synthetic Lieb lattice with two rings, which hence makes a solid step towards experimentally constructing more complicated lattices in multiple rings associated with synthetic frequency dimension.

physics.optics↗

arXiv2021

TableSense: Spreadsheet Table Detection with Convolutional Neural Networks

Spreadsheet table detection is the task of detecting all tables on a given sheet and locating their respective ranges. Automatic table detection is a key enabling technique and an initial step in spreadsheet data intelligence. However, the detection task is challenged by the diversity of table structures and table layouts on the spreadsheet. Considering the analogy between a cell matrix as spreadsheet and a pixel matrix as image, and encouraged by the successful application of Convolutional Neural Networks (CNN) in computer vision, we have developed TableSense, a novel end-to-end framework for spreadsheet table detection. First, we devise an effective cell featurization scheme to better leverage the rich information in each cell; second, we develop an enhanced convolutional neural network model for table detection to meet the domain-specific requirement on precise table boundary detection; third, we propose an effective uncertainty metric to guide an active learning based smart sampling algorithm, which enables the efficient build-up of a training dataset with 22,176 tables on 10,220 sheets with broad coverage of diverse table structures and layouts. Our evaluation shows that TableSense is highly effective with 91.3\% recall and 86.5\% precision in EoB-2 metric, a significant improvement over both the current detection algorithm that are used in commodity spreadsheet tools and state-of-the-art convolutional neural networks in computer vision.

cs.IR↗

arXiv2021

NameRec*: Highly Accurate and Fine-grained Person Name Recognition

In this paper, we introduce the NameRec* task, which aims to do highly accurate and fine-grained person name recognition. Traditional Named Entity Recognition models have good performance in recognising well-formed person names from text with consistent and complete syntax, such as news articles. However, there are rapidly growing scenarios where sentences are of incomplete syntax and names are in various forms such as user-generated contents and academic homepages. To address person name recognition in this context, we propose a fine-grained annotation scheme based on anthroponymy. To take full advantage of the fine-grained annotations, we propose a Co-guided Neural Network (CogNN) for person name recognition. CogNN fully explores the intra-sentence context and rich training signals of name forms. To better utilize the inter-sentence context and implicit relations, which are extremely essential for recognizing person names in long documents, we further propose an Inter-sentence BERT Model (IsBERT). IsBERT has an overlapped input processor, and an inter-sentence encoder with bidirectional overlapped contextual embedding learning and multi-hop inference mechanisms. To derive benefit from different documents with a diverse abundance of context, we propose an advanced Adaptive Inter-sentence BERT Model (Ada-IsBERT) to dynamically adjust the inter-sentence overlapping ratio to different documents. We conduct extensive experiments to demonstrate the superiority of the proposed methods on both academic homepages and news articles.

cs.CL↗

arXiv2021

CelebA-Spoof Challenge 2020 on Face Anti-Spoofing: Methods and Results

As facial interaction systems are prevalently deployed, security and reliability of these systems become a critical issue, with substantial research efforts devoted. Among them, face anti-spoofing emerges as an important area, whose objective is to identify whether a presented face is live or spoof. Recently, a large-scale face anti-spoofing dataset, CelebA-Spoof which comprised of 625,537 pictures of 10,177 subjects has been released. It is the largest face anti-spoofing dataset in terms of the numbers of the data and the subjects. This paper reports methods and results in the CelebA-Spoof Challenge 2020 on Face AntiSpoofing which employs the CelebA-Spoof dataset. The model evaluation is conducted online on the hidden test set. A total of 134 participants registered for the competition, and 19 teams made valid submissions. We will analyze the top ranked solutions and present some discussion on future work directions.

cs.CV↗

arXiv2019

Preliminary demonstration of flexible dual-energy X-ray phase-contrast imaging

Currently, dual-energy X-ray phase contrast imaging is usually conducted with an X-ray Talbot-Lau interferometer. However, in this system, the two adopted energy spectra have to be chosen carefully in order to match well with the phase grating. For example, the accelerating voltages of the X-ray tube are supposed to be respectively set as 40 kV and 70 kV, with other energy spectra being practically unusable for dual energy imaging. This system thus has low flexibility and maneuverability in practical applications. In this work, dual energy X-ray phase-contrast imaging is performed in a grating-based non-interferometric imaging system rather than in a Talbot-Lau interferometer. The advantage of this system is that, theoretically speaking, any two separated energy spectra can be utilized to perform dual energy X-ray phase-contrast imaging. The preliminary experimental results show that dual-energy X-ray phase contrast imaging is successfully performed when the accelerating voltages of the X-ray tube are successively set as 40 kV and 50 kV. Our work increases the flexibility and maneuverability when employing dual-energy X-ray phase-contrast imaging in medical diagnoses and nondestructive tests.

physics.med-ph↗

arXiv2018

Motionless and fast measurement technique for obtaining the spectral diffraction efficiencies of a grating

The measurement of the spectral diffraction efficiencies of a diffraction grating is essential for improving the manufacturing technique and for assessing the grating's function in practical applications. The drawback of the currently popular measurement technique is its slow speed due to the hundreds of repetitions of two kinds of time-consuming mechanical movements during the measuring process (i.e., the rotation of the mechanical arm to capture the light beam, and the mechanical variation of the output wavelength of the grating monochromator). This limitation greatly restricts the usage of this technique in dynamic measurement. In this manuscript, we present a motionless and fast measurement technique for obtaining the spectral diffraction efficiencies of a plane grating, effectively eliminating the aforementioned two kinds of mechanical movement. Herein, the proposed solution for removing the first kind of mechanical movement is tested, and the experimental result shows that the proposed method can be successfully used to measure the plane transmission grating's spectral diffraction efficiencies in the wavelength range of 550-750 nm. The method for eliminating the second kind of mechanical movement is not verified in this manuscript; however, we think that it is very straightforward and commercially available. We estimate that the spectral measurement can be achieved on a millisecond timescale by combining the two solutions. Our motionless and fast measuring technique will find broad applications in dynamic measurement environments and mass industrial testing.

physics.ins-det↗

arXiv2018

Apparatus for measuring the uniformity of the optical transmittance of a semispherical surface at normal incidence

In this paper, we build an apparatus for measuring the optical transmittance and its uniformity for a semispherical surface at normal incidence; the system is primarily comprised of a traditional double-beam photometric framework and a novel custom-made mechanical structure with multidimensional degrees of freedom. During the measurement process, a key aligning step is adopted to guarantee that the center point of the semispherical surface stands still in the light beam while scanning the hemispherical optical element point by point around the horizontal and vertical axes, which ensures that the laser beam is always normally incident onto the surface of the hemisphere. The experimental results show that the uniformity of the optical transmittance for a semispherical optical glass can be successfully characterized by the system, with a three-cycle repeatability error of 0.026% being demonstrated. Our system solves the problem of traditional spectrophotometers when measuring the spectral property of a hemispherical surface and thus can be popularized in similar applications.

physics.optics↗

arXiv2017

Setup for measuring the optical propagation time in matter

Optical propagation time in matter could reveal fruitful information, such as the velocity of light and the sample's refractive index. In this paper, we build a simple and robust setup for measuring the optical propagation time in matter for a known distance, the system uses high frequency square signal as the signal carrier, and a lock-in amplifier is employed to obtain the phase difference between the reference square signal and the other one penetrating the sample, in this way the optical time of flight in matter can be obtained by a background subtraction process. Primary experimental result confirms the feasibility of the newly proposed measuring theory, which can be used to measure easily in high-speed the speed of light and the refractive index of optical transparent material, compared with the currently popular measuring technique using oscilloscope, potential advantage of our proposed method employing lock-in amplifier is that high accuracy are promising, and in contrast with the presently most popular method for measuring the sample's refractive index based on the minimum deviation angle, superiority of our suggested method is the easy preparation of the sample, the convenient operability and the fast measuring speed.

physics.ins-det↗